I have a n X 2 matrix, but I want remove the points with a distance smaller than a threshold. And the data is random, which means the data position may be very far in the matrix but their distance is very close. Can anyone help me?

I have a n X 2 matrix, but I wanna remove the points with a distance smaller than a threshold. And the data is random, which means the data position may be very far in the matrix but their distance is very close. Can anyone help me with an algorithm to deal with it?

[KSSV]

You want to remove the points based on distance, but distance from which point?

[Jan Orwat]

Distance between datapoints themselves I assume

[binbin cui]

The distance between any two points should be detected.

Answer 1

[Jan Orwat]

dataOut = uniquetol(dataIn, tolerance, 'ByRows', true);

Above code should be a good starting point. In this case it would be 1-norm distance between points against given tolerance. Check uniquetol documentation for more information.

While it is not "Euklidean" 2-norm distance, in many cases should be good enough and it's very simple workaround.

[binbin cui]

Thank you very much. But I just find another algorithm to implement this calculation. I will use the D=pdist() to get all the possible distances.then find the values below the threshold in the D matrix. Use the index of target values to infer their pairing in original matrix.

[Jan Orwat]

Ok